Re: [PATCH v4] xfs: introduce protection for drop nlink

["Darrick J. Wong" <djwong@kernel.org>]

On Mon, Sep 18, 2023 at 02:50:50PM +0800, cheng.lin130@zte.com.cn wrote:
> From: Cheng Lin <cheng.lin130@zte.com.cn>
> 
> When abnormal drop_nlink are detected on the inode,
> return error, to avoid corruption propagation.
> 
> Signed-off-by: Cheng Lin <cheng.lin130@zte.com.cn>
> ---
>  fs/xfs/xfs_inode.c | 8 ++++++++
>  1 file changed, 8 insertions(+)
> 
> diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
> index 9e62cc500..d16cb62ce 100644
> --- a/fs/xfs/xfs_inode.c
> +++ b/fs/xfs/xfs_inode.c
> @@ -919,6 +919,14 @@ xfs_droplink(
>  	xfs_trans_t *tp,
>  	xfs_inode_t *ip)
>  {
> +
> +	if (VFS_I(ip)->i_nlink == 0) {
> +		xfs_alert(ip->i_mount,
> +			  "%s: Deleting inode %llu with no links.",
> +			  __func__, ip->i_ino);

We're not deleting the zero-nlink file, since we return without changing
the link count.

Also wondering why this doesn't use XFS_IS_CORRUPT?  That macro already
logs a message about there being problems:

	struct xfs_mount	*mp = ip->i_mount;

	if (XFS_IS_CORRUPT(mp, VFS_I(ip)->i_nlink == 0))
		return -EFSCORRUPTED;

--D

> +		return -EFSCORRUPTED;
> +	}
> +
>  	xfs_trans_ichgtime(tp, ip, XFS_ICHGTIME_CHG);
> 
>  	drop_nlink(VFS_I(ip));
> -- 
> 2.18.1